Wake on Lan issues (VT6105/6106S[Rhine-III] network card)

4 réponses [Dernière contribution]
nad
nad
Hors ligne
A rejoint: 01/19/2013

Hello.
Personally, I believe it is mighty rude to start your first forum post with a help request, but I'm quite in a predicament so I've got no choice.

First of all, I love the distribution, as someone who came from a Windows/Ubuntu/Crunchbang background, it's amazing that everything the aformentioned did is doable on a completely libre distro, it must have been a lot of work and effort to get everything running, and must I thank you for it.

Well, here is the problem.
I got two network cards on my board, the built-in realtek one(RTL8111/8168B PCI Express Gigabit Ethernet Controller) which is eth0 running the r8169 driver, this info is taken from ifconfig, ip and nm-tool

And the second card: Via Technologies VT6105/6106S[Rhine-III] on eth1 running the via-rhine driver.
This one is a year or so old, amusingly enough I bought it for my old motherboard that got its internal network card broken.
I have no idea why on the retail box it says D-link but the card is from VIA(maybe it's a VIA chip inside but D-link is one of their partners?, oh well)
The weird part is that both have the same hardware/MAC address( so the issue might be that, but I doubt it.

The problem is, the day I installed Trisquel, which was 4 days ago, everything was working fine, and at that time the second network card(VIA) wasn't even in my PC since the built-in one worked fine.

After a day.. I think, and after a restart, I noticed my transmission client just hanging there not connecting to any peers.
I could not connect anywhere, my browser was completely hanging on opening anything, not in the way that it would immediately show you a error message because you're not connected, but the tab throbber of the browser was indefinitely spinning, not only that, apt-get update and ping completely timed out, I could not even ssh into my machine from my laptop.
Which was strange since nm-applet showed that I was connected.
nm-tool also showed that State: connected (see attachment)

That's part 1 of the problem.

After a bit of mucking around, I gave up and got the VIA/D-link card out, naturally I expected it to work out of the box since it said 'supported on Linux' on the box sticker.
Thankfully It did, and that card had no issues until I found out it just didn't want to wake up from suspend with either etherwake or gwakeonlan.
That's the only issue the card has, and yes, the card does support Wake on Magic packet and I was sure to enable WoL with "ethtool -s eth1 wol g "and it was working on a Windows machine.

And that concludes part 2.

Some misc info: on the first day I installed Trisquel, I actually did notice my built in card displaying similar symptoms, after restarting(this was after the big distupgrade-upgrade download that comes after a fresh install), I had to restart because my browser, and other things were completely timing out, a restart fixed it for several hours, then it came back and stayed to this day, after which I got the VIA card out.

I honestly don't expect both problems to be fixed, but either one would be great.
I tried an hour ago, I could wake up my PC when the ethernet cable was connected to the realtek card, but I still can't actually browse or do anything online.
I can browse and do everything normally on the VIA card, but I can't wake it up via magic packet no matter what I do while on Trisquel
I guess I could potentially connect both of them to the router, or keep my PC always on, but that seems inefficient and bad on my power bill.

I would really love to get either issue fixed, and I'd rather not rely on proprietary firmware unless it's absolutely necessary.

Thank you.

Pièce jointeTaille
Screenshot at 2013-01-19 22:37:57.png13.57 Ko
nad
nad
Hors ligne
A rejoint: 01/19/2013

Testing it out a bit more, I just noticed that there seems to be conflict on the MAC address of my VIA card

nm-tool, ip, ifconfig say that AA:00* is the hw address while the connection settings of 'Wired connection 1' under 'Device MAC adress' for eth1 says that 00:22* is the hw address.

For eth0, the address is completely different 90:2B* but I could only wake it up using the AA:OO* address.

I honestly don't know what's going on.
Attachment for a bit of clarification..

Screenshot at 2013-01-19 23:18:46.png
andrew
Hors ligne
A rejoint: 04/19/2012

The MAC address under "Connection Information" is the one being used.

You should be able to temporarily change your MAC address using this (thanks Chris):

sudo ifconfig ethX down
sudo ifconfig ethX hw ether xx:xx:xx:xx:xx:xx
sudo ifconfig ethX up

Both devices *should* be different MAC addresses. No idea what's going on there TBH.

andrew
Hors ligne
A rejoint: 04/19/2012

> I got two network cards on my board, the built-in realtek one(RTL8111/8168B PCI Express Gigabit Ethernet Controller) which is eth0 running the r8169 driver

There are apparently some issues with that driver. See these pages:

* https://bugs.launchpad.net/ubuntu/+source/linux-backports-modules-3.0.0/+bug/839393
* https://bugs.launchpad.net/ubuntu/+source/linux/+bug/998200
* https://bugzilla.kernel.org/show_bug.cgi?id=43270

Apparently the Linux 3.1+ kernel should be a little bit better. You can backport a kernel from one of these:

http://jxself.org/linux-libre/
https://launchpad.net/~linux-libre/+archive/ppa

nad
nad
Hors ligne
A rejoint: 01/19/2013

Thank you so much, updating the kernel fixed the issue with the built-in adapter and everything is working as it should.

I was aprehensive of installing non-distro maintaned kernels, as I always thought X, mesa and other stuff was also required to update but that doesn't seem to be the case.